Course/Your Quiet Setup

6.1

Privacy and What AI Does With Your Words

The information that should stay out.

What you'll leave with

By the end of this lesson, you'll know clearly and confidently which categories of information should not be entered into AI tools — and why.

Why this matters

AI tools process what you type. That processing happens on servers, not just on your device. Before using AI for anything, it is worth knowing what should stay out — not out of paranoia, but out of basic information hygiene.

The list is shorter than people expect. Most things are fine to use. A specific category of things is not.

The idea

A practical test: if you would not put this information on a shared Google document where strangers could see it, do not put it into AI.

You can describe a situation involving sensitive information without including the sensitive information itself. AI can help you with the language without the data.

The teaching block

Information to keep out:

  • Passwords or security credentials of any kind
  • Financial account numbers, credit card numbers, or banking details
  • Social security or national identification numbers
  • Full medical records or highly sensitive health information
  • Other people's private information without their knowledge
  • Confidential work information that is covered by an NDA or company policy
  • Personal details about third parties who have not consented

For workplace use: many companies have AI policies that specify what can and cannot be processed through external AI tools. If you are unsure whether yours does, ask. Using AI for work without knowing the policy is a real professional risk.

Example

Show the difference between sharing too much and sharing enough.

Too much:

My account number is [XXXX]. There was an unauthorised charge on [date] for $[amount]. Please help me write a dispute letter.

Enough:

Please help me write a dispute letter to my bank about an unauthorised charge. I want to be clear, professional, and include a request for immediate investigation. [leave placeholders for specific details I will fill in]

Both produce a useful output. One exposes sensitive data unnecessarily.

Try this now

Review the prompts you have written during this course. Is there anything you included that falls into the categories above?

Going forward: before you write a prompt, take five seconds to ask — is there sensitive information here I can remove or replace with a placeholder?

Save this

If you would not put this information on a shared Google document where strangers could see it, do not put it into AI.

Quiet takeaway

Most casual AI use involves no sensitive information and carries minimal privacy risk. The goal is not paranoia — just a brief five-second check before you type.

Next

What not to include: covered. Next: the specific considerations for using AI at work, where the stakes and the rules are different.

Sign in to track progress